view release on metacpan or  search on metacpan

lib/Bio/Tradis/CombinePlots.pm  view on Meta::CPAN

    my @file_hs;
    foreach my $f (@files){
        open(my $fh, "<", $f);
        push(@file_hs, $fh);
    }
    return ( id => $id, len => $len, files => \@file_hs );
}

sub _get_file_len {
    my ( $self, $files ) = @_;

    #check all files are of equal lens and return len if true
    #wc misses last line - return $l++
    my @lens;
    for my $f ( @{$files} ) {
        my $wc = `wc $f | awk '{print \$1}'`;
        chomp $wc;
        push( @lens, $wc );
    }

    my $l = shift @lens;
    for my $x (@lens) {
        return 0 if ( $x != $l );
    }
    return $l+1;
}

sub _combine_lines {
    my ( $self, $lines ) = @_;

    my @totals = ( 0, 0 );
    foreach my $l ( @{$lines} ) {
        if(!defined($l)){
            return "";
            next;
        }
        my @cols = split( /\s+/, $l );
        $totals[0] += $cols[0];
        $totals[1] += $cols[1];
    }
    return join( " ", @totals );
}

sub _write_stats_header {
    my ($self) = @_;
    my @fields =
      ( "ID", "Sequence Length", "Unique Insertion Sites", "Seq Len/UIS" );
    print { $self->_stats_handle } join( ",", @fields ) . "\n";
    return 1;
}

sub _write_stats {
    my ( $self, $id, $seq_len ) = @_;
    my $combined_dir = $self->combined_dir;
    my $comb_plot = "$combined_dir/$id.insert_site_plot";

    #my $seq_len = `wc $comb_plot | awk '{print \$1}'`;
    #chomp($seq_len);
    my $uis = `grep -c -v "0 0" $comb_plot`;
    chomp($uis);
    my $sl_per_uis = "NaN";
    $sl_per_uis = $seq_len / $uis if($uis > 0);

    my $stats = "$id,$seq_len,$uis,$sl_per_uis\n";
    print { $self->_stats_handle } $stats;

    return 1;
}

sub _abs_path_list {
    my ( $self, $files ) = @_;
    my $plot_path = $self->_get_plotfile_path;

    my @pathlist;
    foreach my $f ( @{$files} ) {
        if   ( $f =~ /^\// ) { push( @pathlist, $f ); }
        else                 { push( @pathlist, $plot_path . $f ); }
    }
    return \@pathlist;
}

sub _get_plotfile_path {
    my ($self) = @_;
    my $plotfile = $self->plotfile;

    my @dirs = split( '/', $plotfile );
    pop(@dirs);
    my $path2plot = join( '/', @dirs );
    return "$path2plot/";
}

sub _is_gz {
    my ( $self, $plotname ) = @_;

    if ( $plotname =~ /\.gz$/ ) {
        return 1;
    }
    else {
        return 0;
    }
}

sub _unzip_plots {
    my ( $self, $files ) = @_;
    my $destination_directory = $self->_destination;

    my @filelist = @{ $self->_abs_path_list($files) };
    my @unz_plots;
    foreach my $plotname ( @filelist ) {
        Bio::Tradis::Analysis::Exceptions::FileNotFound->throw("Cannot find $plotname\n") unless ( -e $plotname );
        if ( $self->_is_gz($plotname) ) {
            $plotname =~ /([^\/]+$)/;
            my $unz = $1;
            $unz =~ s/\.gz//;
            my $unzip_cmd = "gunzip -c $plotname > $destination_directory/$unz";
            system($unzip_cmd);
            push(@unz_plots, "$destination_directory/$unz");
        }
        else {
            push(@unz_plots, $plotname);
        }